Reuters reports the European Commission has issued antitrust charges against Meta over a policy that blocks rival AI services from using the WhatsApp Business API, and it is weighing interim measures to prevent ‘serious and irreparable’ harm to competition while the case proceeds. The practical effect is that API access rules for business messaging are being treated as competition infrastructure for AI assistants.
Reuters says Meta has pushed back publicly, arguing the WhatsApp Business API is not a key distribution channel for AI chatbots and that users can access AI tools through many other routes. This is the core factual dispute that will matter for any interim order, how gateway status is evidenced and how quickly harm is shown.
Capgemini announces a “sovereign-ready” cloud and AI offer built on AWS Sovereign Cloud for European enterprises, framed explicitly around sovereignty and regulatory expectations. The governance implication is that architecture as compliance is becoming a mainstream procurement narrative, especially where data residency and control of access pathways are sensitive.
Regulation
The European Commission has notified Meta of possible interim measures and sent a Statement of Objections, setting out its preliminary view that Meta may have breached EU antitrust rules by excluding third-party AI assistants from accessing and interacting with users via WhatsApp. The Commission frames the risk as foreclosure in the fast-growing AI assistants market, and interim measures are positioned as a way to prevent serious harm to competition while the investigation continues.
The ICO states it has fined MediaLab.AI, owner of Imgur, £247,590 for children’s privacy failures, including no effective age checks and unlawful processing of under-13s’ data without parental consent, plus failure to carry out a DPIA. This is a direct compliance signal for AI systems relying on user-generated content and profiling.
Cases
- Clean Up The Internet states it has launched a legal challenge against Ofcom’s refusal to disclose basic information about its engagement with major technology companies, and says the case is listed to be heard by the First-tier Tribunal (General Regulatory Chamber) on 26 February 2026. The governance relevance is procedural and institutional: it tests transparency expectations around regulator–platform interaction in the online safety regime. The ICO’s FOIA decision notice (IC-403651-J1L1) records that Ofcom was entitled to withhold some requested information under FOIA section 44 and to neither confirm nor deny holding other information under section 44(2), in response to requests seeking meeting dates, agendas and attendees involving major platforms and relevant lobbying bodies.
Academia
arXiv posted “Machine Learning Practitioners’ Views on Data Quality…” (Feb 2026), which is governance-relevant because it ties day-to-day practitioner reality to legal data governance duties under GDPR and the EU AI Act, helping translate “data governance” into actionable controls.
IAPP argues proposed “Digital Omnibus” changes intended to ease GDPR friction for AI training may miss the mark, focusing on legal basis questions for personal data use in model training and the complexity around special-category data. This is useful for compliance teams because it maps the direction of travel and the contested points likely to shape enforcement and guidance.
Events
RE WORK announces Chief AI Officer Summit UK, 25 February 2026, London, focused on enterprise AI delivery under governance pressure, with sessions framed around production deployment, strategy, and operational challenges.
IAPP publishes details for UK Intensive 2026 in London, 23-26 February 2026, positioned around privacy practice with explicit coverage of AI governance and current UK compliance developments. It is a practitioner-heavy programme that typically focuses on operational controls rather than theory.
Takeaway
AI governance is being enforced through two levers that move faster than abstract principles. First, competition scrutiny is landing on distribution gates like messaging APIs where product rules can decide which assistants reach users. Second, enterprises are increasingly buying “sovereign” deployment patterns as a compliance answer, making architecture decisions part of the governance record.
Sources: Reuters, European Commission, Capgemini, INFORRM, IAPP, ICO, arXiv, cleanuptheinternet